Despite that fact that all humans potentially have perceptual access to the world and have ways of linguistically expressing these perceptions, the question remains how sensory domains are carved in each language. What are the specific linguistic resources available such that some percepts are ineffable?

This paper argues that the productive morphology of Yucatec Maya, a language spoken in Southern Mexico, provides speakers with linguistic resources to talk about specific sensory perceptions. In Yucatec Maya, the lexicon is divided into two classes: a noun class and a verbo-nominal root class. A large portion of verbo-nominal roots can be derived (using reduplication and special suffixes) according to a common template that encodes particular perceptual features of the world (e.g. tak’, ‘adhere(nce)’ becomes tak’-lemak ‘sticky’, chak-tak’-e’en, ‘dirty red’, etc.). Some derivations fit some roots more readily than others according to their semantics, nonetheless these derivations provide a default pattern for speakers to express particular perceptual modalities (namely sight and touch) and specific properties of percepts (mainly agency, completeness, texture, colour and spatial distribution). These derivational processes raise questions about the core meaning of the verbo-nominal roots that seem to encode a skeleton concept (e.g. roundedness, piled-upedness), rather than concrete properties (e.g. a ball or bulge, to stack or to pile up). This linguistic system has consequences for the language-culture interface. Contrary to previous claims, Yucatec Maya suggests that there is no causal relationship between the size of the colour or texture lexicon and the particulars of the environment or material culture.